JOURNAL OF GEOPHYSICAL RESEARCH: PLANETS, VOL. 118, 2030–2044, doi:10.1002/jgre.20157, 2013

Insights into the subsurface structure of the Caloris basin, Mercury, from assessments of mechanical layering and changes in long-wavelength topography Christian Klimczak,1 Carolyn M. Ernst,2 Paul K. Byrne,1 Sean C. Solomon,1,3 Thomas R. Watters,4 Scott L. Murchie,2 Frank Preusker,5 and Jeffrey A. Balcerski 6 Received 15 June 2013; revised 7 September 2013; accepted 11 September 2013; published 3 October 2013.

[1] The volcanic plains that fill the Caloris basin, the largest recognized impact basin on Mercury, are deformed by many graben and wrinkle ridges, among which the multitude of radial graben of Pantheon Fossae allow us to resolve variations in the depth extent of associated faulting. Displacement profiles and displacement-to-length scaling both indicate that faults near the basin center are confined to a ~ 4-km-thick mechanical layer, whereas faults far from the center penetrate more deeply. The fault scaling also indicates that the graben formed in mechanically strong material, which we identify with dry basalt-like plains. These plains were also affected by changes in long-wavelength topography, including undulations with wavelengths of up to 1300 km and amplitudes of 2.5 to 3 km. Geographic correlation of the depth extent of faulting with topographic variations allows a first-order interpretation of the subsurface structure and mechanical stratigraphy in the basin. Further, crosscutting and superposition relationships among plains, faults, craters, and topography indicate that development of long-wavelength topographic variations followed plains emplacement, faulting, and much of the cratering within the Caloris basin. As several examples of these topographic undulations are also found outside the basin, our results on the scale, structural style, and relative timing of the topographic changes have regional applicability and may be the surface expression of global-scale interior processes on Mercury. Citation: Klimczak, C., C. M. Ernst, P. K. Byrne, S. C. Solomon, T. R. Watters, S. L. Murchie, F. Preusker, and J. A. Balcerski (2013), Insights into the subsurface structure of the Caloris basin, Mercury, from assessments of mechanical layering and changes in long-wavelength topography, J. Geophys. Res. Planets, 118, 2030–2044, doi:10.1002/jgre.20157.

1. Introduction Guest,1975;Murchie et al., 2008; Watters et al., 2009a], which, after their emplacement, were subjected to extensional [2] The Caloris basin, with a diameter of ~1550 km the and contractional brittle deformation as well as changes in largest well-preserved impact basin on Mercury (Figure 1a) long-wavelength topography [Zuber et al., 2012]. The brittle [Murchie et al., 2008], preserves a complex tectonic history. fi deformation is evident by the juxtaposition of normal- and The interior of the basin is lled with smooth volcanic plains thrust-fault-related landforms [Strom et al., 1975; Melosh units [Murray et al., 1975; Strom et al., 1975; Trask and and McKinnon,1988;Watters et al., 2005, 2009a; Murchie et al., 2008]. Additional supporting information may be found in the online version of [3] The Mercury Dual Imaging System (MDIS) [Hawkins this article. et al., 2007] on the MErcury Surface, Space ENvironment, 1 Department of Terrestrial Magnetism, Carnegie Institution of Washington, GEochemistry, and Ranging (MESSENGER) spacecraft ac- Washington, D.C., USA. fi 2The Johns Hopkins University Applied Physics Laboratory, Laurel, quired the rst full images of the Caloris basin during its initial Maryland, USA. Mercury flyby in 2008. MDIS data revealed that the basin’s 3Lamont-Doherty Earth Observatory, Columbia University, Palisades, interior smooth volcanic plains units have spectral properties New York, USA. 4 distinct from those of plains units exterior to the basin Center for Earth and Planetary Studies, National Air and Space [Murchie et al., 2008; Robinson et al., 2008]. Further, spectral Museum, Smithsonian Institution, Washington, D.C., USA. 5German Aerospace Center, Institute of Planetary Research, Berlin, differences within the Caloris smooth plains were found to be Germany. present in ejecta blankets, floors, and central peaks of several 6Department of Earth, Environmental, and Planetary Sciences, Case impact craters, suggesting that those impacts had excavated Western Reserve University, Cleveland, Ohio, USA. distinct material from depth [Murchie et al., 2008; Ernst Corresponding author: C. Klimczak, Department of Terrestrial Magnetism, et al., 2010]. From estimates of the excavation depths of spec- Carnegie Institution of Washington, 5241 Broad Branch Road, N.W., trally distinct materials exposed in crater deposits, Ernst et al. Washington, DC 20015-1305, USA. ([email protected]) [2010] derived a stratigraphy for several plains-covered areas ©2013. American Geophysical Union. All Rights Reserved. on Mercury and concluded that the interior plains in the central 2169-9097/13/10.1002/jgre.20157 part of the Caloris basin are between 2.5 and 4 km thick.

2030 KLIMCZAK ET AL.: TECTONICS OF THE CALORIS BASIN

Figure 1. Overview of the Caloris basin area. (a) Location of the Caloris basin (black outline) relative to major expanses of smooth plains on Mercury (shown in blue, modified from Denevi et al. [2013]). (b) Digital terrain model (DTM) of the topography [Preusker et al., 2011] of the Caloris basin and the region to its west overlaid on a Mercury Dual Imaging System (MDIS) monochrome base map, in north polar stereographic projection centered at 140°E. Indicated in white are craters that show a notable tilt of their floors; arrows indicate the direction of tilt, but arrow length does not scale to tilt amount. The systematic tilts of crater floors away from topographic highs constrain the timing of long-wavelength topographic undulations and assist in the mapping of crests and troughs. M = Mozart basin; R = Raditladi basin. The location of the profile in Figure 1c is shown as a black, partially dashed line. (c) Profile A–A′ shows that topographic undulations can be approximately represented by a sinusoidal curve with a wavelength of 1300 km and an amplitude of 3 km.

[4] Stereophotogrammetric analysis of MESSENGER flyby The first MESSENGER flyby revealed graben of basin-radial images yielded the first topographic information for the Caloris and -concentric orientations [Murchie et al., 2008], including basin [Oberst et al., 2010] and revealed long-wavelength the prominent complex of radial graben in the center of the basin, undulations of the basin floor (Figure 1b). Orbital observations now named Pantheon Fossae. There is no consensus yet for the with MESSENGER’s Mercury Laser Altimeter (MLA) instru- origin of Pantheon Fossae [Head et al., 2008, 2009; Freed et al., ment [Cavanaugh et al., 2007] confirmed these long-wavelength 2009; Watters et al., 2009a; Klimczak et al., 2010], but the fault topographic undulations [Zuber et al., 2012] and documented geometric analysis of Klimczak et al. [2010] suggested that the that the floors of several young, flat-floored impact craters graben-bounding faults are vertically confinedtostratigraphic within and around the basin are not horizontal but instead or mechanical units in the central Caloris basin. tilt in the same direction as the regional trend of the long- [6] Orbital images with substantially improved lighting wavelength topography (Figure 1). These observations sug- geometry and resolution (Figures 2 and 3) now permit an in- gest that topographic changes affected the area in and outside depth mechanical analysis of the faults of the Pantheon the basin after the emplacement of the plains, with parts of the Fossae graben. Such fault analysis can improve the character- basin interior units displaced to elevations greater than those ization of the depth extent of faulting throughout the Caloris of the basin rim. basin and can test whether there is a vertical confinement of [5] After their emplacement, the Caloris interior smooth the graben in the basin center. Information on the depth extent plains were affected by extensional and compressional tectonic of faulting and the confinement of the faults to mechanical or stresses, manifest by the many normal- and thrust-fault-related stratigraphic layering provides insight into the emplacement of landforms throughout the basin (i.e., graben and wrinkle the interior volcanic plains as well as the timing and processes ridges). Images returned by the Mariner 10 spacecraft, which involved in the changes in long-wavelength topography across viewed less than half of the basin, showed wrinkle ridges with the region. In this paper we characterize the shapes of the orientations both radial and concentric to the basin center, as graben-bounding faults within the Caloris basin in terms of well as graben of no preferred orientation that outline polygo- their lengths and displacements, infer the depth extent of nal blocks in the plains [Strom et al., 1975; Melosh and faulting, and compare these results to the topography within McKinnon,1988;Thomas et al., 1988; Watters et al., 2005]. and surrounding the Caloris basin.

2031 KLIMCZAK ET AL.: TECTONICS OF THE CALORIS BASIN

Figure 2. MDIS monochrome base map of the Caloris basin interior (250 m/pixel). Note the prominent basin-radial graben (Pantheon Fossae) in the center of the basin, and the basin-circumferential graben far- ther from the center. Also, note the many thrust-fault-related landforms throughout Caloris. The graben used for the displacement analysis shown in Figure 4 are numbered and shown in yellow. The outline of Figure 3 is shown in red. This map is in orthographic projection centered at 30°N, 162°E.

2. Faults in the Caloris Basin We focused our analysis on the description and characteriza- tion of selected graben of Pantheon Fossae. In particular, we [7] Whereas brittle deformation exterior to the Caloris studied individual graben-bounding faults by investigating basin is evident mainly as individual lobate scarps and wrin- the displacement (slip) distribution along the fault, and kle ridges, arrays of such features, or wrinkle-ridge-ringed we statistically evaluated the full graben population by ghost craters that host graben without a preferred orientation examining the scaling behavior of fault displacements with [Klimczak et al., 2012; Watters et al., 2012], faulting within respect to their lengths. We used results from this analysis the basin is manifest as a complex and systematic pattern of to derive information on the depth extent of faulting crosscutting graben and wrinkle ridges [e.g., Byrne et al., throughout the basin by extrapolation to other faults within 2013]. Both wrinkle ridges and graben occur in radial and this population. circumferential orientations with respect to the basin center (Figure 2). Throughout the basin, wrinkle ridges have mostly circumferential orientations, whereas the graben toward the 2.1. Fault Restriction in the Central Caloris Basin center are found to be exclusively radial (Figures 2 and 3). [9] Klimczak et al. [2010] inferred a nearly constant maxi- With improved lighting geometries from MESSENGER mum depth of faulting of ~3 km for the faults that bound a orbital imaging, radial graben are observed to be more numer- substantial number of graben in the central Caloris basin, ous and continue beyond what was originally described and and they suggested that those faults may have been vertically mapped for Pantheon Fossae [Head et al., 2009; Watters confined to a mechanical or stratigraphic layer. Layering in a et al., 2009a; Klimczak et al., 2010], to radial distances of up faulted geological medium can have a marked influence on to 500 km from the basin center. Beyond that point, these the dimensions of faults and their growth and scaling behav- structures are less common, and circumferential graben and ior [Soliva et al., 2005]. Thus, layering in the Caloris basin graben lacking a preferred orientation dominate (Figure 2). should be resolvable from a careful characterization of the Ridges in the outer part of the Caloris basin are also circumfer- fault dimensions. ential or of no preferred orientation (Figure 2). [10] Fault dimensions include the fault length (L) in the [8] The abundance of faults in the Caloris basin provides along-strike direction, as well as the width or height (H), a basis for assessing the rock-mechanical and geometric measured in the downdip direction, and the average fault properties of the volcanic plains unit in which they formed. displacement (D). Information from at least two of these

2032 KLIMCZAK ET AL.: TECTONICS OF THE CALORIS BASIN

fault aspect ratio [Schultz and Fossen, 2002]. The fault length is found to scale in proportion to the maximum displacement (Dmax)[Cowie and Scholz, 1992a, 1992b; Scholz et al., 1993; Schultz and Fossen, 2002; Schultz et al., 2010] as:

Dmax ¼ γ L; (1)

where γ is a scaling coefficient related to rock-mechanical properties, the fault aspect ratio, and the driving stress condi- tions acting on the faults of a given population [Schultz and Fossen, 2002; Schultz et al., 2006]. Such fault growth behav- ior leads to a constant accumulation of displacement, with displacement minima at the fault tips and a maximum near the center of the fault, resulting in an overall peaked along- strike distribution of displacement [Cowie and Scholz, 1992b, 1992c; Dawers et al., 1993; Manighetti et al., 2001; Scholz, 2002]. [11] In contrast, vertical confinement of faults within a single stratum or mechanical layer limits fault growth and the further accumulation of displacement in the downdip direction. As a result, sufficiently large faults increase their lengths but not their heights [Soliva et al., 2005; Soliva and Benedicto, 2005; Soliva et al., 2006]. Such faults no longer grow in a self-similar manner, i.e., fault aspect ratios and thus eccentricities of the elliptical fault shape increase continu- ously as faults lengthen. This behavior results in a nonlinear displacement-to-length scaling relationship [Schultz and Fossen, 2002; Soliva et al., 2005; Soliva and Benedicto,2005;Soliva et al., 2006; Polit et al., 2009], and a flat-topped (i.e., plateau- shaped) along-strike displacement distribution [Dawers et al., 1993; Manighetti et al., 2001; Soliva et al., 2005; Soliva and Benedicto, 2005; Soliva et al., 2006; Politetal., 2009]. Although nonlinear displacement-to-length scaling and plateau- shaped displacement distributions can also arise from fault interaction and linkage [e.g., Cartwright et al., 1995; Wyrick et al., 2011], such fault growth behavior has been widely documented for restricted faults in a variety of terrestrial and planetary settings [Dawers et al., 1993; Soliva et al., 2005; Polit et al., 2009]. 2.1.1. Displacement Distributions [12] We measured the lengths and displacements of 31 Figure 3. Mosaic of a set of MDIS wide-angle camera faults in the Caloris basin (Figure 2) and constructed dis- (WAC) images of the center of the Caloris basin, illustrating placement distribution profiles for each. Fault lengths of the structural complexity of the area. Note that none of the single faults were measured from their mapped linear traces impact craters in this view is crosscut by any of the faults. between the two fault tips identified on MDIS images. Examples of isolated graben are indicated by green arrows, Displacements on surface-breaking normal faults on airless soft-linked graben (relay ramps) are indicated by white arrows, planetary bodies with minimal erosion can be obtained by and hard-linked graben are marked with red arrows. This measuring the relief of the trough-forming graben, which mosaic is in orthographic projection centered at 30°N, 162°E. yields the vertical component of the fault displacement Individual images used to assemble the mosaic include (the fault throw). However, neither stereophotogrammetry- EW0220375075G, EW0220375107G, EW0220375141G, derived topographic data sets nor laser altimetric data for EW0220375183G, EW0220418418G, and EW0220418460G. Mercury are suited for such fine-scale topographic analysis, as those data sets do not have sufficient vertical or horizontal resolution, respectively. Therefore, fault throws were ac- parameters can yield insight into the three-dimensional char- quired by shadow measurements of graben on MDIS mono- acter of the fault surface, which is usually idealized as an chrome images with resolutions of 98–133 m per pixel; we ellipse [Irwin, 1962; Kassir and Sih, 1966; Willemse et al., measured shadow widths in 1 km increments along the entire 1996; Willemse, 1997; Schultz and Fossen, 2002] with an length of the studied faults to obtain the throw distribution of eccentricity defined as the aspect ratio of fault length to fault each. For this analysis, we selected only the best illuminated height (L/H). In an infinite, elastic, homogenous medium, and best preserved faults, as sufficiently accurate measure- faults grow in a self-similar manner, i.e., they have a constant ments could be made only for a limited range of lighting

2033 KLIMCZAK ET AL.: TECTONICS OF THE CALORIS BASIN

Figure 4. Measured distribution of fault throw along the length of several normal faults or fault arrays within the Caloris basin. Vertical exaggeration is 40:1. (a) Throw distribution along two overlapping faults shown in relation to an interpreted sketch map and an MDIS monochrome image. Each data point in the distribution represents one shadow measurement, and error bars denote the measurement uncertainty of ± one pixel. The area in blue represents the distribution of the cumulative throw of the two individual faults. The skew of the maxima in displacement away from the centers of the individual faults and toward the center of the fault array indicates that the two fault segments interacted. (b) A series of throw distribu- tions, shown at the same scale and arranged by decreasing fault length, indicate the dependence of fault throw on fault length. Longer faults have higher throws. Also, note that longer faults have plateau-shaped (i.e., flat-topped) distributions. conditions and only for those faults for which the morphol- detailed description of the measurement procedure and ogy was not later modified by impact cratering or mass assessment of sources of error.) wasting. Shadows were measured by pixel counting. Errors [13] A representative set of fault throw profiles, sorted by on the order of half a pixel (~50–65 m) are equivalent fault length, is shown in Figure 4, and a high-resolution view to measurement uncertainties of ± 10 m in the vertical of these faults can be found in the supporting information. direction. MLA profiles, where available, compare well Profiles include fault throw data from arrays of two faults to shadow measurements. (See Appendix A for a more (Figure 4, fault arrays 1, 3, 4) as well as from individual faults

2034 KLIMCZAK ET AL.: TECTONICS OF THE CALORIS BASIN

segments. Profiles with such a throw distribution resemble those for normal fault arrays on Earth [Peacock and Sanderson,1994;Cartwright et al., 1995; Dawers and Anders,1995;Willemse et al., 1996; Willemse,1997; Manighetti et al., 2001; Soliva and Benedicto,2004;Soliva et al., 2008] and Mars [Polit et al., 2009] and indicate interac- tion between the individual fault segments. [15] Fault 2 (Figure 4b) is a single, isolated 75-km-long fault with a generally plateau-shaped throw distribution, but with higher throws present in portions of the profile (at ~15 km and ~40 km along strike). Fault arrays 3 and 4 each consist of two individual segments and have total lengths of ~50 km. Although the arrays are in different parts of the Caloris basin (Figure 2), the two profiles look similar to one another and have throw distributions with plateaus along the segment closer to the basin center in each case. Because throw maxima are not located adjacent to the stepover region, individual fault segments likely have not interacted markedly. Isolated faults 5, 6, and 8 are ~30–40 km long, and all have generally flat- topped throw distributions (Figure 4b). Isolated faults 7, 9, and 10 show peaks in their throw distribution (Figure 4b). [16] Throw distribution profiles for faults 2–10 (Figure 4b) Figure 5. Maximum displacements of several planetary nor- highlight the scale dependence of fault displacement accumu- mal fault populations as functions of fault length. Note that lation. Small faults, with lengths shorter than ~30 km, have shorter faults in the individual populations display a linear scal- peaked profiles indicative of self-similar, unrestricted fault ing behavior, whereas longer faults in some populations (i.e., growth. Many of the larger faults in the central Caloris basin, those with a flat-topped distribution of fault throw) deviate those for which lengths exceed ~30 km, in contrast, have flat- from linear scaling, indicating fault restriction. The faults topped throw distribution profiles, indicating nonproportional of Pantheon Fossae show both types of scaling behavior. fault growth (Figure 4b, faults 2–6, 8). We interpret such Restriction curves (black dotted lines) are modeled according growth behavior as evidence for confinement of the longer to the methodology described in Appendix B. The gray-shaded faults to a distinct mechanical layer. The plateaus in the throw linear trend is the modeled best fit for the unrestricted faults distributions for the longer faults therefore developed because of Pantheon Fossae. This best fit is for a semi-elliptical fault these faults, once restricted, grew only in the horizontal direc- length ratio of 0.5L/H = 2. NP = northern plains graben, Mars, tion, so they increased their lengths but did not grow further in restricted to faults with heights of 2–3km[Polit et al., 2009]; the vertical dimension and thus did not accumulate additional AP = Alba Patera, Mars, unrestricted [Polit et al., 2009]; downdip displacement. Plateaus in the throw distribution TT = Tempe Terra, Mars, unrestricted [Polit et al., 2009]; C1 show local peaks (e.g., Figure 4b, fault 2), indicating that the (this study) = Pantheon Fossae faults, Mercury, restricted to fault surface grew to greater depth due to local variations faults with heights of 3–4.5 km (light red) or unrestricted in fault geometry or layer properties. Longer faults in the (red); C2 = faults with no preferred orientation in Caloris, no Caloris basin (those with lengths exceeding 30 km) that have restriction apparent [Watters and Nimmo, 2010]. peaked throw distributions had either unrestricted vertical growth, or penetrated into the underlying rock unit before resuming self-similar growth [Schultz and Fossen,2002; (Figure 4b, faults 2, 5–10). Individual graben consist of Soliva et al., 2005; Soliva and Benedicto, 2005]. isolated fault pairs that show no evidence on MDIS images 2.1.2. Displacement-Length Relations for interaction with other graben (Figure 3, green arrows). [17] Comparison of the maximum displacements and cor- Arrays consist of sets of two or more graben in a closely responding lengths of faults within the same population gives spaced, en echelon geometry (Figure 3, white arrows). There insight into the scaling statistics and growth behavior of indi- is evidence for fault interaction within some of these graben vidual faults in that population. Downdip fault displacements arrays. Knowledge of whether faults interacted with each other can be obtained from fault throws for an assumed dip angle of or linked up is important for interpreting their scaling behavior the fault surface. For a dip angle of 60°, a value consistent and displacement distributions, a topic discussed further in with the optimum dip angle of normal faults in cohesive section 2.1.3. volcanic material [e.g., Jaeger et al., 2007], the approximate [14] Fault array 1 (Figure 4a), an example of a segmented displacements of the faults can be calculated. graben structure in central Caloris, shows the fault throw dis- [18] There is a wealth of literature discussing the scaling rela- tributions of two individual fault segments that overlap tionships of fault displacements to fault lengths [Cailleux, 1958; each other by ~5 km (errors in endpoint locations are ±30 m). Elliott, 1976; Watterson, 1986; Walsh and Watterson, 1988; The fault array is >80 km long and has a maximum throw Gillespie et al., 1992; Cowie and Scholz, 1992b; Dawers of ~500 m. The two segments of this array show peaked pro- et al., 1993; Schlische et al., 1996; Schultz et al., 2008]. More files, with their individual throw minima at the fault tips and recent studies have reached a general consensus on direct pro- maxima offset from the segment center toward the midpoint portionality, i.e., a linear correlation between the two parame- of the whole array, adjacent to the stepover between the ters [equation (1)] [Cowie and Scholz,1992b;Dawers et al.,

2035 KLIMCZAK ET AL.: TECTONICS OF THE CALORIS BASIN

linearly are found farther from the center (C1, red squares). This pattern indicates that only faults in the center of the basin are confined to a mechanical layer, whereas normal faults at greater radial distances from the center grew in an un- restricted, self-similar manner. The presence of restricted and unrestricted faults in different parts of the basin may be a func- tion of variations in thickness and/or properties of the surficial mechanical layer across the basin. [21] Displacement-versus-length data for graben in the Caloris basin with circumferential orientations or with no pre- ferred orientation were derived by Watters and Nimmo [2010] from Mariner 10 images and stereo-derived topography. That data set (C2 in Figure 5), for which individual measurements have uncertainties of ±10 m, is more or less consistent with our measurements but gives a slightly higher scaling coeffi- cient. Differences in the scaling factor may arise from any of a variety of reasons, including fault populations with different length/height aspect ratios, different driving stress conditions, or measurement uncertainties. [22] The values of the Dmax/L ratio of restricted faults can be converted to an estimate of the mechanical thickness of Figure 6. Contours of maximum depth extent of faulting for the layer in which they are restricted. Following the mathe- radial graben of Pantheon Fossae in relation to the topography matical framework for three-dimensional fault scaling of of the Caloris basin. These maximum faulting depths can be Schultz and Fossen [2002], we varied parameters pertaining interpreted as the minimum thickness of the mechanical unit to the elliptical fault shape, host-rock properties, and stress in which the faults grew. Restricted faults are shown as conditions around the fault tip so as to achieve acceptable fits black dots. Note the shoaling of the maximum depth extent of the resultant analytical solution to our Dmax/L measure- of faulting in the central portions of the basin, near the topo- ments. Details of the procedure are outlined in Appendix B. graphic low to the southwest of Apollodorus crater (A), where Our results indicate that fits to our Dmax/L measurements can an increased density of restricted faults is observed. The loca- be achieved only for a narrow range of input parameters. In tions of the profiles (along segments of great circles) shown particular, for a rock unit of mechanical properties consistent in Figure 7 are indicated by black lines. This map is in ortho- with dry basalt, together with shear stress conditions appro- graphic projection centered at 30°N, 162°E. priate for such materials on Mercury, the faults that deviate from linear scaling are restricted to fault heights of 3–4.5 km (Figure 5). For a 60° fault dip angle, this restriction corre- 1993; Schlische et al., 1996; Scholz,2002;Schultz et al., sponds to a vertical confinement of the faults within a mechan- 2008]. In contrast, nonlinear scaling of fault growth is found ical unit of ~2.6–4 km thickness. to be representative of situations in which the fault is confined 2.1.3. Fault-Linkage Effects to a layer with rock-mechanical properties distinct from those [23] Fault linkage occurs when the overlap and spacing of the underlying material [Schultz and Fossen,2002;Soliva between two or more faults are sufficiently close that the et al., 2005; Schultz et al., 2010]. Detailed field measurements faults mutually influence each other’s growth. In map view, confirm such scaling behavior in fault populations that are the linkage of normal faults is evident by ramps in the confined to a single mechanical layer [Soliva et al., 2005; stepover region (relay structures) or by abrupt bends ( jogs) Soliva and Benedicto,2005;Soliva et al., 2006]. in the fault trace. In some cases, fault linkage can have effects [19] Displacement maxima for 36 faults within Pantheon on the Dmax/L relationship and slip distribution profiles Fossae (Figure 2) are presented as a function of fault length that are similar to those produced by fault confinement to a in Figure 5, together with normal fault populations from mechanical or stratigraphic layer. An array of several small, Mars [Polit et al., 2009] and elsewhere on Mercury (faults softly linked faults, i.e., interacting but not fully coalesced near the rim of the Caloris basin studied with Mariner 10 data fault segments [see Schultz, 2000] with substantial overlap, by Watters and Nimmo [2010], data set C2). Whereas data might show flat-topped (i.e., plateau-shaped) slip distribu- sets from both Tempe Terra (TT) and Alba Patera (AP) on tions, and a population of linked faults can show a nonlinear Mars show a largely linear Dmax/L scaling, a data set from Dmax/L scaling relationship [e.g., Cartwright et al., 1995; the northern plains of Mars (NP) shows a systematic decrease Schultz et al., 2010; Wyrick et al., 2011] resembling that of in the slope of the Dmax/L relation with increasing L [Polit restricted faults. Although linked faults are present in the et al., 2009]. This scaling behavior was interpreted [Polit Caloris smooth plains (e.g., Figure 4a), fault throw distribu- et al., 2009] as indicating confinement of faults to a layer tions (Figure 4b) generally show profiles typical of either 2–3 km thick (Figure 5). restricted or unrestricted fault growth [e.g., Manighetti [20] The data set we compiled from Pantheon Fossae shows et al., 2001], and all flat-topped throw distribution profiles both a linear relationship and a deviation from linear scaling show plateaus at around the same maximum displacement (Figure 5, data set C1). Faults that follow the nonlinear (i.e., value of ~180 m (Figure 4b). These combined results indicate restricted) scaling (C1, light red squares) are located mainly that fault linkage has at most a minor influence on our analysis in the central portion of Caloris basin, whereas faults that scale of fault scaling.

2036 KLIMCZAK ET AL.: TECTONICS OF THE CALORIS BASIN

Figure 7. Topographic profiles through the Caloris basin, together with inferred mechanical stratigraphy (the red line shows the maximum depth of faulting; the line is solid where faults are restricted and dashed where faults are unrestricted). The location of the basin rim along each profile is indicated by a black arrow. Also shown with arrows are maximum excavation depths (upward arrow) and minimum depths of origin of central peak material (downward arrow) for large impact craters [Ernst et al., 2010]. Arrows are blue for cra- ters that excavated material spectrally distinct from the surrounding terrain and tan for craters that excavated material spectrally similar to the surface unit. If a nearby crater does not fall directly on the profile, its eleva- tion is projected onto the profile and thus might differ from the topography shown. All profiles are shown with 20:1 vertical exaggeration. (a) Profile B–B′ shows sinusoidal topography of 1120 km wavelength and 2.5 km amplitude. Inferred mechanical stratigraphy shows notable undulations in the depth to the base of the surface mechanical unit; the undulations are generally consistent with spectra of material excavated by large craters and the inferred excavation depths. (b) Profile C–C′ shows sinusoidal topography of 1070 km wavelength and 3 km amplitude. Inferred mechanical stratigraphy shows shoaling toward the basin center, consistent with spectra of crater ejecta and inferred excavation depths. (c) Profile D–D′ shows sinusoidal topography of 850 km wavelength and 2.5 km amplitude. Inferred mechanical stratigraphy shows steady thickening toward the center of the profile. (d) Profile E–E′ shows minor topographic undulations, but they cannot be modeled with simple sinusoidal functions. Inferred mechanical stratigraphy shows shoaling toward the basin center, consistent with spectra of crater ejecta and inferred excavation depths.

2.2. Depth Extent of Faulting however, information on fault height can be used only to [24] The established statistical relationship between maxi- infer the maximum depth extent of faulting. This value can mum displacement and fault length for the structures in then be used to infer the minimum depth extent of the unit Pantheon Fossae (Figure 5) can be used to obtain information in which the faults grew. on fault height and thus the depth extent of faulting for every [25] The three-dimensional fault scaling analysis yielded fault in this population. Where fault growth is restricted, the good fits for unrestricted faults with constant length aspect fault height can be directly related to the depth of a mechan- ratios of 0.5L/H = 2 (see Appendix B for details), such that ical boundary. Where faults grew in an unrestricted manner, the length of the fault is approximately four times the fault

2037 KLIMCZAK ET AL.: TECTONICS OF THE CALORIS BASIN height. From this relationship, and for a 60° fault dip angle, 3. Topographic Undulations in and Around the the depth extent of faulting can be estimated for every radial Caloris Basin fault in the basin. We estimated the depth of faulting of ~150 faults with mostly basin-radial orientations throughout Caloris [29] The topography of the Caloris basin (Figures 1 and 6) (see the supporting information). is characterized by long-wavelength undulations of the basin floor, with some areas having been vertically displaced so as [26] From the data set for maximum depth of faulting to be now above the basin rim [Zuber et al., 2012]. In addi- below the topographic surface in the Caloris basin, we inter- fl fl polated with the “blockmean” and “surface” tools in the open tion, the at portions of oors of several impact craters within geo-processing software GMT [Wessel and Smith, 1998] Caloris are tilted in the same direction as the downslope trend (http://gmt.soest.hawaii.edu/). From this interpolation, we of the long-wavelength topography [Balcerskietal., 2012, obtained a contour map that approximates the maximum 2013; Zuber et al., 2012]. Similar relationships between long- depth of faulting below the present topography (Figure 6). wavelength topography and tilted craters can be observed in The contours, shown in relation to the basin topography the regions exterior to the basin (Figure 1). [Preusker et al., 2011], illustrate that the maximum depth [30] The undulations can be described as gentle, elongate topographic highs and lows, and systematic trends of crater extent of faulting is variable across the basin, with graben- fl bounding faults penetrating from ~4 to more than 7 km below oor tilts support the mapping of long-wavelength troughs the surface. A zone in which the maximum faulting depth is and crests (Figure 1). Within the Caloris basin, there is a <4.5 km, within which faults are inferred to be confined to trough in the basin center and two crests to the north and a mechanical unit on the basis of their displacement/length south of that central topographic low (Figure 1). In the region relationships, is located near the basin center. That zone is west of the basin, a crest can be mapped through the topo- surrounded by an annulus in which faults penetrate to some- graphic high that hosts Raditladi basin, as well as two associ- ated troughs to its north and south (Figure 1). what greater depths (Figure 6). The maximum depth extent of fi faulting lessens toward the basin rim. [31] In cross-strike pro les, the topographic undulations can be approximated by long-wavelength (850–1300 km) but low- [27] From displacement distributions and the analysis of amplitude (2.5–3 km) harmonic functions (Figures 1c and 7). maximum displacement versus fault length, we infer that fi – ′ the majority of faults are confined to a mechanical unit near Pro le A A (Figure 1c) shows undulations outside the fi Caloris basin that match a sinusoidal form with a wavelength the center of Caloris basin, but we nd no evidence for con- fi – ′ – ′ finement outside of the central basin zone. These obser- of 1300 km and an amplitude of 3 km. Pro les B B to E E vations suggest that many of the graben we have analyzed (Figure 7) show that topographic trends within the Caloris are contained within a single coherent mechanical unit. basin match sinusoidal forms of somewhat shorter wave- Under this interpretation, the thickness of this unit correlates lengths (850 to 1120 km) than but similar amplitudes (2.5 to directly with the maximum depth extent of faulting in regions 3 km) to those for undulations outside the basin. The axes of of fault confinement, whereas only minimum estimates of all mapped troughs and crests in and around the Caloris basin are preferentially oriented approximately east–west (Figure 1), the thickness of the mechanical unit may be deduced in – – ′ regions of unconfined fault growth. Contours of maximum but the east west cross section E E (Figure 7d) shows that extent of faulting (Figure 6) may be interpreted, by this some minor topographic variations also occur perpendicular view, as isopachs that represent the minimum thickness of to the orientation of the major axes of the undulations. [32] The presence of topographic undulations inside the the uppermost mechanical layer in the Caloris basin, and fi so imply a mechanical boundary at those depth values or faulted and mechanically strati ed Caloris basin is useful greater depths. Alternatively, regions of greater maximum for gaining insight into the nature of topographic changes depths of faulting may correspond to areas in which faults on Mercury. Combining topographic information with esti- penetrated a mechanical boundary and propagated into the mates of mechanical layering at depth allows us to interpret underlying unit. the subsurface structure of the basin and permits assessment of the processes involved in long-wavelength deformation on [28] The systematic decrease of fault width and length to- ward the edge of the basin for graben with no preferred orienta- Mercury. This assessment provides information on lithologic tion to the basin rim (see supporting information) is suggestive stratigraphy and rock-mechanical properties of the smooth of a decreased penetration depth of the faults. By a similar plains, the origin of the long-wavelength topographic undula- argument, the presence of a wider, younger set of graben in tions, and the timing of topographic changes relative to smooth the central Caloris basin with larger geometric proportions for plains emplacement, smooth plains deformation, and the the individual structures (Figures 2 and 3) has been regarded cratering record within the Caloris basin. as indicative of a greater depth extent of faulting for these faults than the adjacent, narrower structures [Klimczak et al., 4. Discussion 2010]. However, these wider graben are no longer fully preserved because subsequent impacts have modified portions 4.1. Depth Extent of Faulting Versus Mechanical/ of them, and so detailed information on their lengths, slip Lithologic Stratigraphy distributions, and scaling behavior is unreliable for any [33] Correlation of the depth extent of faulting with to- interpretation of the stratigraphy of the rocks beneath the up- pography for four representative profiles across the Caloris permost mechanical layer of smooth plains. Notwithstanding basin shows that faults grew to different depths below the such limitations, these graben display widths of up to ~9 km present surface (Figure 7). The depth extent of faulting in and so they must have penetrated the mechanical boundary the Caloris basin appears to correlate with long-wavelength inferred to be present at ~4 km depth in the center of topography in some locations (e.g., Figure 7c) and to Caloris basin. anticorrelate with long-wavelength topography in other

2038 KLIMCZAK ET AL.: TECTONICS OF THE CALORIS BASIN locations (e.g., Figure 7b). Under our interpretation that the is consistent with stratigraphic relationships inferred from maximum depth of faulting represents the minimum thick- the depth extent of faulting. Cunningham crater (38 km in ness of the uppermost mechanical layer, such relations indi- diameter), in contrast, exposes spectrally distinct material cate variable but systematic thickness changes in the only on its central peak, constraining the change in lithology mechanical stratigraphy of Caloris. For the alternative sce- to a depth between the excavation depth of the ejecta and nario, under which the unrestricted graben-bounding faults that of the central peak. The estimated minimum excavation penetrated the bottom of the upper mechanical layer, a layer depth of the spectrally distinct central peak material of boundary would be discernable only in the center of the Cunningham is somewhat less than the inferred depth of basin, where faults are restricted. the mechanical boundary in the vicinity of the crater, but [34] Profile B–B′ (Figure 7a) in the western part of the as this value is a minimum this observation is not inconsistent Caloris basin (Figure 6) shows a topographic high (crest) to with our findings. the north and a low (trough) to the south. The depth extent [37] Along profile C–C′ (Figure 7b), the depths of excava- of the faults generally increases inward from the basin mar- tion of Poe and Apollodorus craters both show a good corre- gin but appears to shoal somewhat beneath the basin center. lation to the inferred mechanical stratigraphy. Poe crater The depth extent of faulting appears to anticorrelate with (77 km in diameter) displays spectrally distinct material in the long-wavelength topography along profile C–C′ through its ejecta and its central peak, and both associated estimates the center of the basin (Figure 7b); i.e., unrestricted faults of excavation depth exceed the depth of the inferred mechan- penetrate deeper beneath high-standing terrain and fault ical boundary. Apollodorus crater (41 km in diameter), in restriction indicative of a mechanical contrast occurs beneath contrast, displays spectrally distinct material only on its long-wavelength lows near the basin center. Profile D–D′ central peak. The estimated minimum excavation depth of through the eastern portion of the basin shows a steady central peak material for Apollodorus exceeds the inferred increase of the depth extent of unrestricted faulting toward depth of the mechanical boundary, which in this area is given the center (Figure 7c). The east–west–oriented profile E–E′ by the depth of confinement of the graben-bounding faults. (Figure 7d) shows fault restriction near the center of the Along profile E–E′ (Figure 7d), the depths of excavation Caloris basin and unrestricted faults that penetrate deeper of Cunningham, Atget, and an unnamed crater (marked away from the center. as “1”) also correlate with mechanical stratigraphy. For [35] Many large impact craters are found on the smooth Cunningham crater, as for Apollodorus along profile C–C′, plains within the Caloris basin (Figure 2). Several of the only the minimum depth of excavation of central peak mate- larger impact craters have exposed spectrally distinct mate- rial exceeds the inferred depth of the mechanical boundary, rials in their ejecta or on their floors and central peaks [e.g., showing a good fit between changes in lithology and Ernst et al., 2010]. Spectral variations among the major color mechanical properties inferred from faulting depths. Atget units on Mercury are generally attributed to compositional crater (100 km in diameter), in contrast, exposes spectrally differences [Robinson et al., 2008; Blewett et al., 2009]. distinct material on its floor as well as its central peak, and Different spectral properties of material exposed in the ejecta the estimated depths of excavation for material in both parts or central peaks of impact craters may therefore yield addi- of the crater exceed the inferred depth of the mechanical tional information on the lithological stratigraphy within boundary. The unnamed crater (31 km in diameter) shows Caloris. Ernst et al. [2010] estimated the depth of excavation very little spectrally distinct material on its central peak. Its of spectrally distinct materials associated with large craters in estimated minimum depth of excavation of central peak the central Caloris basin. These authors concluded, on the material is somewhat shallower than the inferred mechanical basis of the diameters of several craters that expose spec- boundary, but because portions of the central peak are not trally distinct materials on their central peaks but not on spectrally distinct from the ejecta and surrounding terrain their floors or ejecta, that the high-reflectance red plains for this crater, and because its excavation depth estimate is that partially fill the central Caloris basin must generally be be- a minimum, these results do not conflict with our findings. tween 2.5 and 4 km thick. This range is remarkably similar to [38] The broad agreement between the excavation depths that for the depth of confinement of fault growth (Figure 5) in of spectrally distinct material and the depth of the mechanical the central portion of the basin (i.e., 2.6–4km). boundary beneath the Caloris basin interior inferred from the [36] We have compared estimates of the depth of excava- maximum depth extent of faulting suggests that the change tion of spectrally distinct material by several impact craters in spectral properties at depth beneath the interior smooth within the Caloris basin (on the basis of the results of Ernst plains coincides with the base of the upper mechanical layer. et al. [2010] and shown in the supporting information) with Moreover, results of our three-dimensional fault scaling anal- the mechanical stratigraphy inferred from the fault analysis ysis suggest that the graben formed in a mechanically strong, of this study along the topographic profiles in Figure 7. dry basaltic rock mass (see Appendix B for details). These Estimates can be made for the maximum depth of excavation physical rock properties are consistent with geochemical data of materials in the ejecta and crater floor as well as for the obtained by MESSENGER’s X-Ray Spectrometer (XRS), minimum depth of origin of the central peak (see Ernst et al. which, on the basis of elemental surface abundances, indicate [2010] and references therein). Profile B–B′ (Figure 7a) a basalt-like composition for the smooth plains within the incorporates depth information from three large craters: Caloris basin interior [Weider et al., 2012]. The consistency Sander, Munch, and Cunningham. Sander and Munch cra- between mechanical, spectral, and geochemical properties ters (with diameters of 47 and 57 km, respectively) show for these plains suggests that the mechanical unit in which material spectrally distinct from surrounding terrain in their the graben formed relates lithologically to a ~4-km-thick ejecta, floor, and central peaks, so a stratigraphic boundary smooth plains layer of basaltic composition that is composi- must overlie the maximum excavation depth. This finding tionally distinct from the underlying material.

2039 KLIMCZAK ET AL.: TECTONICS OF THE CALORIS BASIN

rise [e.g., Zuber et al., 2012] located near the center of the northern volcanic plains hosts ghost craters that all tilt away from the crest of the rise, whereas many of the fresher craters there have horizontal crater floors [Klimczak et al., 2012]. This pattern indicates that there, too, the process responsible for the topographic changes was active after plains emplacement. For both the northern and Caloris interior plains, Balcerski et al. [2013] categorized the tilted craters that superpose the plains by state of degradation and found that changes in topography occurred relatively recently in Mercury’s geologic history. Information on the start and duration of changes in long- wavelength topography is more challenging to obtain, how- ever, and requires detailed characterization of crater ages in Figure 8. Variations in crustal thickness [Smith et al., relation to measured floor tilts. 2012] in and around the Caloris basin (black outline). The [42] Prior to the availability of global topographic data sets, area shown corresponds to that in Figure 1a, and the positions long-wavelength topographic undulations were predicted to of axes of topographic crests and troughs correspond to those be present on Mercury as lithospheric folds [Hauck et al., in Figure 1b. Note the tendency for axes of crests and troughs 2004; Dombard and Hauck, 2008]. Such folds were suggested to coincide with zones of thicker and thinner crust, respec- to have been one manifestation of the radial contraction of tively. This map is in north polar stereographic projection Mercury as its interior cooled, because estimates of global centered at 140°E. contractional strain and radius change derived from the esti- mated horizontal shortening accommodated by the thrust faults that underlie lobate scarps [Strom et al., 1975; Watters 4.2. Tectonic Evolution of the Caloris Basin and Global- et al., 1998, 2009b; Watters and Nimmo, 2010] were far lower Scale Implications than predictions from thermal history models [Hauck et al., [39] The crosscutting and superposition relationships among 2004; Dombard and Hauck, 2008]. faults, impact craters, and topography allow a reconstruction [43] The horizontal contractional strain, ε,acrosstheundu- of the relative timing of the events that led to the present lations on Mercury inferred from the observed wavelengths morphological and topographic characteristics of the Caloris and amplitudes of the undulations, however, is on the order basin. The infill consisted at least of an upper smooth plains of 105 or less. Strains of such low magnitude would contrib- unit that is ~ 4 km thick and is compositionally different from ute insignificantly to the average radial shortening from global underlying units. The lack of evident impact craters partially contraction. Furthermore, numerical modeling of elastic fold- or fully buried by interior smooth plains material indicates ing generally suggests that buckling of a thick lithosphere is either that any younger craters on the Caloris basin floor were difficult to achieve under likely stress conditions [McAdoo deeply buried or that plains emplacement followed basin and Sandwell,1985;Turcotte and Schubert, 2002; Bland formation so closely that there was insufficient time for large and McKinnon, 2012]. Nonetheless, numerical simulations impact craters to form. have shown that buckling of a lithosphere with elastic-plastic [40] After their emplacement, the Caloris smooth plains [McAdoo and Sandwell, 1985] or visco-plastic [Zuber, 1987; were widely deformed by extensional and contractional Montési and Zuber, 2003] deformational behavior can occur tectonic structures, and a multitude of impact craters were without unrealistically high stress magnitudes. superposed on the plains (Figure 2). No fault-related land- [44] Topographic undulations could also result from uneven forms, either graben or wrinkle ridges, appear to crosscut vertical loading of the top of the lithosphere. Indeed, the area any of the impact craters (Figure 3). This result indicates that in and around the Caloris basin was affected by extensive the majority of both extensional and contractional faulting flood volcanism [e.g., Denevi et al., 2013] and the emplace- occurred within the relatively short time span between plains ments of large volcanic loads inside and surrounding the emplacement and the formation of the earliest of the large im- basin. However, topographic undulations occur on a broader pact craters visible today. That the flat portions of the floors spatial scale than the volcanic units and have deformed to of many of the larger craters in the Caloris basin are tilted the same extent areas that have been volcanically buried and [Balcerski et al., 2012; Zuber et al., 2012] (Figure 1) shows other locations that have not. Moreover, these undulations that the process that produced the topographic changes was have wavelengths and amplitudes that are too nearly harmonic active after the formation of the brittle deformational features to reflect the irregular distribution of volcanic units [Denevi and at least some of the subsequent cratering of the Caloris et al., 2013] as sources of lithospheric loading. Therefore, interior plains. we conclude that lithospheric loading is not likely to have been [41] Long-wavelength topographic changes can be detected the dominant source of the topographic undulations. globally on Mercury, as there is evidence for such changes [45] Long-wavelength topographic undulations might also both outside the Caloris basin (Figure 1) and elsewhere on be the result of dynamic topography produced by mantle the planet [Klimczak et al., 2012; Solomon et al., 2012; convective processes. Numerical models show that mantle Zuber et al., 2012]. Comparison between the different regions convection may have occurred on Mercury [King, 2008; that were affected by topographic changes and their strati- Roberts and Barnouin, 2012; Michel et al., 2013] despite a graphic relationships to volcanic plains units, faults, and cra- mantle thickness of less than 400 km [Smith et al., 2012; ters may provide insights into the timing and duration of the Hauck et al., 2013]. The axes of the crests and troughs of process that led to these changes. For example, a topographic the topographic undulations (Figure 1b) show some

2040 KLIMCZAK ET AL.: TECTONICS OF THE CALORIS BASIN correlation with crustal thickness variations derived from a depth of excavation of exposed materials in impact craters model of Mercury’s gravity field expanded to spherical har- that show spectral differences from the surface plains unit monic degree and order 20 [Smith et al., 2012] (Figure 8). [Ernst et al., 2010] correlate well with our inferred mechani- Axes of troughs largely coincide with thinner regions of the cal stratigraphy, supporting the inference that changes in rock crust, whereas axes of crests mainly collocate with areas of mechanical properties generally coincide with variations in thicker crust. Because the wavelengths of model artifacts rock spectral properties. for the current harmonic gravity field solution are on the or- [49] Furthermore, characterization of the long-wavelength der of those of the topographic undulations, however, confir- topography suggests that the area in and around the Caloris mation of this finding and further geophysical interpretation basin has been affected by topographic changes, with elon- will require solutions for Mercury’s gravity field to substan- gate high-standing and low-lying regions corresponding to tially higher harmonic degree and order. the crests and troughs of sinusoidal forms when viewed in [46] The orientations of the graben and wrinkle ridges cross section (Figure 7). Several of the cross sections through appear largely unrelated to the long-wavelength changes of the Caloris basin indicate that the thickness of the uppermost topography within the basin. This lack of correlation is consis- mechanical unit of the Caloris smooth plains varies systemat- tent both with the inference that the faults formed as products ically with topography, suggesting a complex subsurface of basin-localized tectonics well before the topographic structure of the basin. 5 changes as well as with the low strains (ε ~10 )accommo- [50] The stratigraphic and topographic properties of the dated by the undulations. Although numerical results are sup- Caloris plains, coupled with crosscutting and superposition re- portive of the idea that faulting may have accompanied the lationships among plains, craters, and faults, yield insight into formation of lithospheric-scale undulations [Neumann and the relative timing of events in the basin. In particular, the pro- Zuber,1995;Montési and Zuber, 2003], we infer that stresses cesses responsible for the topographic changes were active associated with the formation of the topographic undulations well after both plains emplacement and most brittle deforma- on Mercury were too small to have exerted a large effect on tion inside Caloris. The stratigraphic and temporal relation- preexisting or younger brittle structures. ships of topographic undulations have global implications, as [47] The presence of topographic undulations on Mercury, these features occur elsewhere on Mercury and thus may be and their stratigraphic relationship with other landforms in the surface manifestations of such interior processes as global the Caloris basin, not only allows for better characterization contraction or mantle convection. of the geologic and tectonic evolution of the basin but also has implications for dynamic processes and the thermal evolu- Appendix A: Error Assessment tion of the planet’s interior. Global contraction or mantle con- vection as candidate processes for the topographic changes [51] Shadows within graben were measured on individu- both suggest that the observed undulations may involve the ally processed MDIS images to maintain the highest possible entire lithosphere. If so, wavelengths and amplitudes of the resolution (~98–133 m per pixel). Measurements were taken undulations may hold useful information regarding the rheo- from orthographic projections (centered on each studied logical properties of the crust and mantle during the time of graben structure to minimize parallax errors) by counting their formation, as rheological contrasts have a major impact the pixels perpendicular to the trace of a graben-bounding on the geometry of the undulations. For example, theory of fault. By accounting for solar incidence angle and graben ori- large-scale buckling of a planetary lithosphere [Ricard and entation relative to solar azimuth, shadow measurements can Froidevaux, 1986; Montési and Zuber, 2003] shows that its be converted to the topographic difference between the crest mechanical structure controls the wavelength of deformation of the graben and its floor (i.e., the fault throw). Optimal [Montési and Zuber, 2003]. Moreover, if these long-wavelength lighting conditions are for solar incidence angles of 60–80°; undulations are indeed related to mantle dynamic processes within this range, we found many graben for which shadows that occurred late in Mercury’s history, a timing suggested were sufficiently wide to be resolved across several pixels by the numerous tilted craters [Balcerski et al., 2013], such but were not so wide that the shadow covered the entire a result would imply that convection cells of sizes compara- graben floor. ble to the half-wavelength of the undulations characterized [52] The vertical distance between graben crest and floor, Mercury’s interior on a similar timeframe. corresponding to the fault throw, was derived from trigono- metric relationships between solar azimuth and incidence angles and the measured shadow width. Measurement uncer- 5. Conclusions tainties are on the subpixel scale. Errors in our shadow width [48] By means of a detailed fault displacement analysis, we measurements can be assumed to be on the order of half a have established the maximum depth extent of faulting for pixel (i.e., approximately 50–65 m). This error reduces to the Pantheon Fossae radial graben complex in the Caloris ba- an uncertainty of ± 10 m in the vertical direction, so each sin. Our results show that in the center of the basin, graben- estimate of fault throw can be assumed to deviate by less than bounding faults are likely confined to a mechanical layer, ~20 m from the actual topographic difference between crests consistent with previous studies [Klimczak et al., 2010]. and floors of the graben. For example, a shadow width of Farther from the basin center, normal faults grew in an 1500 ± 65 m across a graben that is illuminated at a solar unconfined manner or penetrated the mechanical unit. azimuth of 75° with a solar incidence angle of 80° yields a Under the first of these interpretations, the maximum depth fault throw of 220 ± 14 m. A strong correlation between our extent of faulting for this latter group of structures can be measurements and throws derived from MLA profiles across interpreted as representing the minimum thickness of the some of our measured graben (Figure A1) shows that throw mechanical unit in which the faults grew. Estimates for the estimates derived from shadow measurements are robust.

2041 KLIMCZAK ET AL.: TECTONICS OF THE CALORIS BASIN

2003]. The angle from the center of the fault plane to a point on its tip, θ [Irwin,1962;Kassir and Sih, 1966], is equal to zero when displacements are analyzed along the horizontal axis of the fault, i.e., its length, so that the term in equation (A1) that is a function of θ reduces to unity. [55] For crustal scales, values for Young’smodulusofafrac- tured rock volume (rock mass) on a planetary surface are best approximated with the deformation modulus [Bieniawski, 1989; Schultz, 1996], and the modulus increases with depth [Rubin, 1990] as a function of the planetary gravitational acceleration and rock density [Schultz et al., 2006]. We assumed a dry basaltic rock mass of density ρ = 3.1 g/m3, yielding E = 24 GPa at and near the surface of Mercury, a value consistent with considerations by Schultz et al. [2006]. Lower densities result for less competent volcanic rock masses [Bieniawski,1989],asrecentlyconfirmed for surface units on the Moon [Wieczorek et al., 2013], with values for the deformation modulus also lower. Values of the deformation modulus increase only slowly with depth on Mercury, due to Figure A1. Fault throws estimated from shadow mea- the comparatively low surface gravitational acceleration. ’ surements compared with the throws obtained for the Poisson s ratio was assigned a typical value for rock of ν same faults from MLA profiles. Shadow measurements = 0.25. The unfaulted, dry, basaltic rock mass in the Caloris compare well with MLA data, indicating that any system- basin is expected to be mechanically strong, so values of σ atic errors in this analysis are small. The straight line has a y were taken to be 15 ± 5 MPa, on the basis of estimates slope of unity. obtained with the Hoek-Brown failure law [Hoek and Brown,1980;Schultz et al., 2006; Jaeger et al., 2007]. Values for the ratio of yield strengths to driving stresses are [53] Despite minimal erosion on Mercury, some sources of found to be approximately 2–3[Cowie and Scholz, 1992a; uncertainty exist for the interpretation of topographic offsets Schultz and Fossen,2002;Wilkins and Schultz, 2005]. as actual fault throws, because cratering and mass wasting [56] The parameters pertaining to rock properties and may have modified topography in places. Furthermore, stress conditions are expected to remain generally constant the graben floors were assumed to be horizontal for the anal- throughout a given rock mass, so that the solution for ysis, but asymmetric graben geometries and tilted graben equation (A1) depends mainly on the shape of the faults. floors would introduce some error in the throw estimates. Following Polit et al. [2009], the fault shape was parameter- Converting throw to displacement could introduce further ized by the fault aspect ratio, i.e., the ratio of fault length to uncertainty, as fault dip angles were assumed to be uniform fault height L/H for buried faults and by the ratio 0.5L/H at 60°. These uncertainties affect absolute fault displace- for surface-breaking faults (i.e., half the fault ellipse). The ments. Data trends, however, such as the shape of the normal faults in the Caloris basin are clearly surface break- distribution of fault throws, or the slope of a population of ing, so we utilized the latter ratio for our analysis. A linear Dmax versus L, should be little affected, as most uncertainties solution is obtained for constant aspect ratios. Best fits to are systematic and would affect all measurements more or the data along the linear portion of the relation are achieved less equally. with a constant aspect ratio of 2 (Figure 5), consistent with the range of fault aspect ratios on other terrestrial planets Appendix B: Three-Dimensional Fault Scaling [Schultz et al., 2006]. By holding fault heights constant but increasing their lengths, the nonlinear analytical solution [54] For a fault of horizontal length L =2a and downdip was calculated (Figure 5). When the nonlinear trends of the height H =2b, the three-dimensional scaling relationship for measurements and analytical solution match, it can be in- the shear strain accommodated on the elliptical fault surface ferred that the continued growth of individual fault structures fi can be further speci ed from equation (1) as: beyond the range over which the linear portion of the relation nohi holds was restricted to that constant fault height. π σd 2 σd σy 1 cos D max 21ðÞ ν 2 σy ¼ γ ¼ rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiqffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; L E ÀÁ ÀÁ: cos2 θ þ b 2 sin2 θ 1 þ 1:464 a 1 65 [57] Acknowledgments. We thank the associate editor and two anony- a b mous reviewers for helpful comments. CK also thanks Richard A. Schultz fi (A1) for sharing previously published work on con ned fault growth. The MESSENGER project is supported by the NASA Discovery Program under where ν, E, and σ are the Poisson’s ratio, Young’s modulus, contracts NASW-00002 to the Carnegie Institution of Washington and y NAS5-97271 to The Johns Hopkins University Applied Physics Laboratory. and yield strength at the fault tip, respectively, of the host rock [Schultz and Fossen, 2002; Schultz et al., 2006]. The References parameter σd is the cumulative driving stress, accounting for the total number of slip events necessary to produce the Balcerski,J.A.,S.A.HauckII,P.Sun,C.Klimczak,P.K.Byrne,A.J.Dombard, O. S. Barnouin, M. T. Zuber, R. J. Phillips, and S. C. Solomon (2012), Tilted observed fault lengths and displacements [Cowie and Scholz, crater floors: Recording the history of Mercury’s long-wavelength deforma- 1992a; Scholz,1997;Gupta and Scholz,2000;Schultz, tion, Lunar Planet. Sci., 43, abstract 1850.

2042 KLIMCZAK ET AL.: TECTONICS OF THE CALORIS BASIN

Balcerski, J. A., S. A. Hauck II, P. Sun, C. Klimczak, P. K. Byrne, Jaeger, J., N. G. Cook, and R. Zimmerman (2007), Fundamentals of Rock R. J. Phillips, and S. C. Solomon (2013), New constraints on timing and Mechanics, 4th ed., 593 pp., Blackwell, Oxford, U. K. mechanisms of regional tectonism from Mercury’s tilted craters, Lunar Kassir, M. K., and G. C. Sih (1966), Three-dimensional stress distribution Planet. Sci., 44, abstract 2444. around an elliptical crack under arbitrary loadings, J. Appl. Mech., 33, Bieniawski, Z. T. (1989), Engineering Rock Mass Classifications, 251 pp, 601–611. Wiley, New York. King, S. D. (2008), Pattern of lobate scarps on Mercury’s surface reproduced Bland, M. T., and W. B. McKinnon (2012), Forming Europa’s folds: Strain by a model of mantle convection, Nat. Geosci., 1, 229–232. requirements for the production of large-amplitude deformation, Icarus, Klimczak, C., R. A. Schultz, and A. L. Nahm (2010), Evaluation of the 221, 694–709. origin hypotheses of Pantheon Fossae, central Caloris basin, Mercury, Blewett, D. T., M. S. Robinson, B. W. Denevi, J. J. Gillis-Davis, J. W. Head, Icarus, 209, 262–270, doi:10.1016/j.icarus.2010.04.014. S. C. Solomon, G. M. Holsclaw, and W. E. McClintock (2009), Klimczak, C., T. R. Watters, C. M. Ernst, A. M. Freed, P. K. Byrne, Multispectral images of Mercury from the first MESSENGER flyby: S. C. Solomon, D. M. Blair, and J. W. Head (2012), Deformation associ- Analysis of global and regional color trends, Earth Planet. Sci. Lett., 285, ated with ghost craters and basins in volcanic smooth plains on Mercury: 272–282, doi:10.1016/j.epsl.2009.02.021. Strain analysis and implications for plains evolution, J. Geophys. Res., Byrne, P. K., C. Klimczak, D. M. Blair, S. Ferrari, S. C. Solomon, 117, E00L03, doi:10.1029/2012JE004100. A. M. Freed, T. R. Watters, and S. L. Murchie (2013), Tectonic complex- Manighetti, I., G. C. P. King, and Y. Gaudemer (2001), Slip accumulation ity within volcanically infilled craters and basins on Mercury, Lunar and lateral propagation of active normal faults in Afar, J. Geophys. Res., Planet. Sci., 44, abstract 1261. 106, 13,667–13,696. Cailleux, A. (1958), Etude quantitative de failles, Rev. Géomorphologie McAdoo, D. C., and D. T. Sandwell (1985), Folding of oceanic lithosphere, Dynamique, 9, 129–145. J. Geophys. Res., 90, 8563–8569, doi:10.1029/JB090iB10p08563. Cartwright, J. A., B. D. Trudgill, and C. S. Mansfield (1995), Fault growth Melosh, H. J., and W. B. McKinnon (1988), The tectonics of Mercury, by segment linkage: An explanation for scatter in maximum displacement in Mercury,editedbyF.Vilas,C.R.Chapman,andM.S.Matthews, and trace length data from the Canyonlands Grabens of SE Utah, J. Struct. pp. 374–400, University of Arizona Press, Tucson, Ariz. Geol., 17, 1319–1326, doi:10.1016/0191-8141(95)00033-A. Michel, N. C., S. A. Hauck II, S. C. Solomon, R. J. Phillips, J. H. Roberts, and Cavanaugh, J. F., et al. (2007), The Mercury Laser Altimeter instrument for M. T. Zuber (2013), Thermal evolution of Mercury as constrained by the MESSENGER mission, Space Sci. Rev., 131, 451–479, doi:10.1007/ MESSENGER observations, J. Geophys. Res. Planets, 118, 1033–1044, s11214-007-9273-4. doi:10.1002/jgre.20049. Cowie, P. A., and C. H. Scholz (1992a), Displacement-length scaling Montési, L. G. J., and M. T. Zuber (2003), Spacing of faults at the scale of the relationship for faults: Data synthesis and discussion, J. Struct. Geol., 14, lithosphere and localization instability: 1. Theory, J. Geophys. Res., 1149–1156. 108(B2), 2110, doi:10.1029/2002JB001923. Cowie, P. A., and C. H. Scholz (1992b), Growth of faults by accumulation of Murchie, S. L., et al. (2008), Geology of the Caloris basin, Mercury: A view seismic slip, J. Geophys. Res., 97, 11,085–11,095. from MESSENGER, Science, 321,73–76, doi:10.1126/science.1159261. Cowie, P. A., and C. H. Scholz (1992c), Physical explanation for the Murray, B. C., R. G. Strom, N. J. Trask, and D. E. Gault (1975), Surface displacement-length relationship of faults using a post-yield fracture history of Mercury: Implications for terrestrial planets, J. Geophys. Res., mechanics model, J. Struct. Geol., 14, 1133–1148, doi:10.1016/0191- 80, 2508–2514. 8141(92)90065-5. Neumann, G. A., and M. T. Zuber (1995), A continuum approach to the Dawers, N. H., and M. H. Anders (1995), Displacement-length scaling and fault development of normal faults, in Rock Mechanics: Proceedings of the linkage, J. Struct. Geol., 17, 607–614, doi:10.1016/0191-8141(94)00091-D. 35th U.S. Symposium on Rock Mechanics,editedbyJ.K.Daemenand Dawers, N. H., M. H. Anders, and C. H. Scholz (1993), Growth of normal R. A. Schultz, pp. 191–198, Balkema Publishers, Rotterdam, The Netherlands. faults: Displacement-length scaling, Geology, 21, 1107–1110. Oberst, J., F. Preusker, R. J. Phillips, T. R. Watters, J. W. Head, M. T. Zuber, Denevi, B. W., et al. (2013), The distribution and origin of smooth plains on and S. C. Solomon (2010), The morphology of Mercury’s Caloris basin as Mercury, J. Geophys. Res. Planets, 118,891–907, doi:10.1002/jgre.20075. seen in MESSENGER stereo topographic models, Icarus, 209, 230–238, Dombard, A. J., and S. A. Hauck II (2008), Despinning plus global contrac- doi:10.1016/j.icarus.2010.03.009. tion and the orientation of lobate scarps on Mercury: Predictions for Peacock, D. C. P., and D. J. Sanderson (1994), Geometry and development MESSENGER, Icarus, 198, 274–276, doi:10.1016/j.icarus.2008.06.008. of relay ramps in normal fault systems, Amer. Assoc. Petrol. Geol. Bull., Elliott, D. (1976), The energy balance and deformation mechanisms of thrust 78, 147–165. sheets, Phil. Trans. Roy. Soc. London, Ser. A, 283, 289–312. Polit, A. T., R. A. Schultz, and R. Soliva (2009), Geometry, displacement- Ernst, C. M., S. L. Murchie, O. S. Barnouin, M. S. Robinson, B. W. Denevi, length scaling, and extensional strain of normal faults on Mars with inferences D. T. Blewett, J. W. Head, N. R. Izenberg, S. C. Solomon, and on mechanical stratigraphy of the Martian crust, J. Struct. Geol., 31, J. H. Roberts (2010), Exposure of spectrally distinct material by impact 662–673, doi:10.1016/j.jsg.2009.03.016. craters on Mercury: Implications for global stratigraphy, Icarus, 209, Preusker,F.,J.Oberst,J.W.Head,T.R.Watters,M.S.Robinson,M.T.Zuber, 210–223, doi:10.1016/j.icarus.2010.05.022. and S. C. Solomon (2011), Stereo topographic models of Mercury after three Freed, A. M., S. C. Solomon, T. R. Watters, R. J. Phillips, and M. T. Zuber MESSENGER flybys, Planet. Space Sci., 59, 1910–1917, doi:10.1016/j. (2009), Could Pantheon Fossae be the result of the Apollodorus crater- pss.2011.07.005. forming impact within the Caloris basin, Mercury?, Earth Planet. Sci. Ricard, Y., and C. Froidevaux (1986), Stretching instabilities and lithospheric Lett., 285, 320–327, doi:10.1016/j.epsl.2009.02.038. boudinage, J. Geophys. Res., 91,8314–8324. Gillespie, P., J. J. Walsh, and J. Watterson (1992), Limitations of dimension Roberts, J. H., and O. S. Barnouin (2012), The effect of the Caloris impact on and displacement data from single faults and the consequences for data the mantle dynamics and volcanism of Mercury, J. Geophys. Res., 117, analysis and interpretation, J. Struct. Geol., 14, 1157–1172. E02007, doi:10.1029/2011JE003876. Gupta, A., and C. H. Scholz (2000), A model of normal fault interaction Robinson, M. S., et al. (2008), Reflectance and color variations on Mercury: based on observations and theory, J. Struct. Geol., 22, 865–879, Regolith processes and compositional heterogeneity, Science, 321,66–69, doi:10.1016/S0191-8141(00)00011-0. doi:10.1126/science.1160080. Hauck, S. A., II, A. J. Dombard, R. J. Phillips, and S. C. Solomon (2004), Rubin, A. M. (1990), A comparison of rift-zone tectonics in Iceland and Internal and tectonic evolution of Mercury, Earth Planet. Sci. Lett., 222, Hawaii, Bull. Volcanol., 52, 302–319, doi:10.1007/BF00304101. 713–728, doi:10.1016/j.epsl.2004.03.037. Schlische, R. W., S. S. Young, R. V. Ackermann, and A. Gupta (1996), Hauck, S. A., II, et al. (2013), The curious case of Mercury’s internal Geometry and scaling relations of a population of very small rift-related structure, J. Geophys. Res. Planets, 118, 1204–1220, doi:10.1002/ normal faults, Geology, 24, 683–686. jgre.20091. Scholz, C. H. (1997), Earthquake and fault populations and the calculation of Hawkins, S. E., III, et al. (2007), The Mercury Dual Imaging System on the brittle strain, Geowissenschaften, 15, 124–130. MESSENGER spacecraft, Space Sci. Rev., 131, 247–338, doi:10.1007/ Scholz, C. H. (2002), The Mechanics of Earthquakes and Faulting, 471 pp., s11214-007-9266-3. Cambridge University Press, Cambridge. Head, J. W., et al. (2008), Volcanism on Mercury: Evidence from the first Scholz, C. H., N. H. Dawers, J. Z. Yu, M. H. Anders, and P. A. Cowie MESSENGER flyby, Science, 321,69–72, doi:10.1126/science.1159256. (1993), Fault growth and fault scaling laws: Preliminary results, Head, J. W., et al. (2009), Evidence for intrusive activity on Mercury from J. Geophys. Res., 98, 21,951–21,961. the first MESSENGER flyby, Earth Planet. Sci. Lett., 285, 251–262, Schultz, R. A. (1996), Relative scale and the strength and deformability doi:10.1016/j.epsl.2009.03.008. of rock masses, J. Struct. Geol., 18, 1139–1149, doi:10.1016/0191-8141 Hoek, E., and E. T. Brown (1980), Empirical strength criterion for rock (96)00045-4. masses, J. Geotech. Eng. Div. Am. Soc. Civil Eng., 106, 1030–1035. Schultz, R. A. (2000), Fault-population statistics at the Valles Marineris Irwin, G. R. (1962), Crack-extension force for a part-through crack in a plate, extensional province, Mars: Implications for segment linkage, crustal strains, J. Appl. Mech., 29, 651–654. and its geodynamical development, Tectonophysics, 316,169–193.

2043 KLIMCZAK ET AL.: TECTONICS OF THE CALORIS BASIN

Schultz, R. A. (2003), A method to relate initial elastic stress to fault Watters, T. R., and F. Nimmo (2010), The tectonics of Mercury, in Planetary population strains, Geophys. Res. Lett., 30(11), 1593, doi:10.1029/ Tectonics,editedbyT.R.WattersandR.A.Schultz,pp.15–80, Cambridge 2002GL016681. University Press, Cambridge. Schultz, R. A., and H. Fossen (2002), Displacement-length scaling in three Watters, T. R., M. S. Robinson, and A. C. Cook (1998), Topography of dimensions: The importance of aspect ratio and application to deformation lobate scarps on Mercury: New constraints on the planet’s contraction, bands, J. Struct. Geol., 24, 1389–1411. Geology, 26, 991–994. Schultz, R. A., C. H. Okubo, and S. Wilkins (2006), Displacement-length Watters, T. R., F. Nimmo, and M. S. Robinson (2005), Extensional troughs scaling relations for faults on the terrestrial planets, J. Struct. Geol., 28, in the Caloris basin of Mercury: Evidence of lateral crustal flow, Geology, 2182–2193, doi:10.1016/j.jsg.2006.03.034. 33, 669–672, doi:10.1130/G21678.1. Schultz, R. A., R. Soliva, H. Fossen, C. H. Okubo, and D. M. Reeves (2008), Watters, T. R., S. C. Solomon, M. S. Robinson, J. W. Head, S. L. André, Dependence of displacement–length scaling relations for fractures and S. A. Hauck II, and S. L. Murchie (2009a), The tectonics of Mercury: deformation bands on the volumetric changes across them, J. Struct. The view after MESSENGER’s first flyby, Earth Planet. Sci. Lett., 285, Geol., 30, 1405–1411, doi:10.1016/j.jsg.2008.08.001. 283–296, doi:10.1016/j.epsl.2009.01.025. Schultz, R. A., R. Soliva, C. H. Okubo, and D. Mege (2010), Fault Watters, T. R., S. L. Murchie, M. S. Robinson, S. C. Solomon, B. W. Denevi, populations, in Planetary Tectonics, edited by T. R. Watters and R. A. S. L. André, and J. W. Head (2009b), Emplacement and tectonic deforma- Schultz, pp. 457–510, Cambridge University Press, Cambridge. tion of smooth plains in the Caloris basin, Mercury, Earth Planet. Sci. Smith, D. E., et al. (2012), Gravity field and internal structure of Mercury from Lett., 285, 309–319, doi:10.1016/j.epsl.2009.03.040. MESSENGER, Science, 336,214–217, doi:10.1126/science.1218809. Watters, T. R., S. C. Solomon, C. Klimczak, A. M. Freed, J. W. Head, Soliva, R., and A. Benedicto (2004), A linkage criterion for segmented normal C. M. Ernst, D. M. Blair, T. A. Goudge, and P. K. Byrne (2012), faults, J. Struct. Geol., 26,2251–2267, doi:10.1016/j.jsg.2004.06.008. Extension and contraction within volcanically buried impact craters and Soliva, R., and A. Benedicto (2005), Geometry, scaling relations and spacing basins on Mercury, Geology, 40, 1123–1126, doi:10.1130/G33725.1. of vertically restricted normal faults, J. Struct. Geol., 27, 317–325, Watterson, J. (1986), Fault dimensions, displacements and growth, Pure doi:10.1016/j.jsg.2004.08.010. Appl. Geophys., 124, 365–373, doi:10.1007/BF00875732. Soliva, R., R. A. Schultz, and A. Benedicto (2005), Three-dimensional Weider, S. Z., L. R. Nittler, R. D. Starr, T. J. McCoy, K. R. Stockstill-Cahill, displacement-length scaling and maximum dimension of normal faults P. K. Byrne, B. W. Denevi, J. W. Head, and S. C. Solomon (2012), in layered rocks, Geophys. Res. Lett., 32, L16302, doi:10.1029/ Chemical heterogeneity on Mercury’s surface revealed by the 2005GL023007. MESSENGER X-Ray Spectrometer, J. Geophys. Res., 117, E00L05, Soliva, R., A. Benedicto, and L. Maerten (2006), Spacing and linkage of doi:10.1029/2012JE004153. confined normal faults: Importance of mechanical thickness, J. Geophys. Wessel, P., and W. H. Smith (1998), New, improved version of Generic Res., 111, B01402, doi:10.1029/2004JB003507. Mapping Tools released, Eos Trans. Am. Geophys. Un., 79(47), 579. Soliva, R., A. Benedicto, R. A. Schultz, L. Maerten, and L. Micarelli Wieczorek, M. A., et al. (2013), The crust of the Moon as seen by GRAIL, (2008), Displacement and interaction of normal fault segments Science, 339, 671–675, doi:10.1126/science.1231530. branched at depth: Implications for fault growth and potential earth- Wilkins, S. J., and R. A. Schultz (2005), 3D cohesive end-zone model for quake rupture size, J. Struct. Geol., 30, 1288–1299, doi:10.1016/j. source scaling of strike-slip interplate earthquakes, Bull. Seismol. Soc. jsg.2008.07.005. Am., 95,2232–2258, doi:10.1785/0120020106. Solomon, S. C., C. Klimczak, P. K. Byrne, S. A. Hauck II, J. A. Balcerski, Willemse, E. J. M. (1997), Segmented normal faults: Correspondence A. J. Dombard, M. T. Zuber, D. E. Smith, R. J. Phillips, J. W. Head, and between three-dimensional mechanical models and field data, J. Geophys. T. R. Watters (2012), Long-wavelength topographic change on Mercury: Res., 102,675–692. Evidence and mechanisms, Lunar Planet. Sci., 43, abstract 1578. Willemse, E. J. M., D. D. Pollard, and A. Aydin (1996), Three-dimensional Strom, R. G., N. J. Trask, and J. E. Guest (1975), Tectonism and volcanism analyses of slip distributions on normal fault arrays with consequences for on Mercury, J. Geophys. Res., 80, 2478–2507. fault scaling, J. Struct. Geol., 18,295–309, doi:10.1016/S0191-8141(96) Thomas, P. G., P. Masson, and L. Fleitout (1988), Tectonic history of 80051-4. Mercury, in Mercury, edited by F. Vilas, C. R. Chapman, and M. S. Wyrick, D. Y., A. P. Morris, and A. A. Ferrill (2011), Normal fault growth in Matthews, pp. 401–428, University of Arizona Press, Tucson, Ariz. analog models and on Mars, Icarus, 212, 559–567, doi:10.1016/j. Trask, N. J., and J. E. Guest (1975), Preliminary geologic terrain map of icarus.2011.01.011. Mercury, J. Geophys. Res., 80, 2461–2477. Zuber, M. T. (1987), Compression of oceanic lithosphere: An analysis of Turcotte, D. L., and G. Schubert (2002), Geodynamics, 2nd ed., 456 pp. intraplate deformation in the Central Indian Basin, J. Geophys. Res., 92, Cambridge University Press, Cambridge. 4817–4825. Walsh, J. J., and J. Watterson (1988), Analysis of the relationship between Zuber, M. T., et al. (2012), Topography of the northern hemisphere of displacements and dimensions of faults, J. Struct. Geol., 10, 239–247, Mercury from MESSENGER laser altimetry, Science, 336, 217–220, doi:10.1016/0191-8141(88)90057-0. doi:10.1126/science.1218805.

2044